International Journal ofEnvironmental Researchand Public HealthArticleFactors Impacting Risk Perception under TyphoonDisaster in Macao SAR, ChinaYajing Shen 1, Shiyan Lou 2,*, Xiujuan Zhao 3,4,*, Kuai Peng Ip 5, Hui Xu 1 and Jingwen Zhang 11 Institute of Analytical Psychology, City University of Macao, Macao 999078, China;jstzsyj0401@163.com (Y.S.); dahui2008.love@163.com (H.X.); h18092200174@cityu.mo (J.Z.)2 Faculty of Finance, City University of Macao, Macao 999078, China3 Department of Engineering Physics, Tsinghua University, Beijing 100084, China4 Institute of Public Safety Research, Tsinghua University, Beijing 100084, China5 Research Center for Macao Social and Economic Development, City University of Macao, Macao 999078,China; kpip@cityu.mo* Correspondence: sylou@cityu.mo (S.L.); xjzhao2009@163.com (X.Z.)Received: 17 August 2020; Accepted: 2 October 2020; Published: 9 October 2020Abstract: Studying typhoon risk perception and its influencing factors help reveal potential riskfactors from the perspective of the public and provide a basis for decision-making for reducing therisk of typhoon disasters. The purpose of this study is to assess the risk perception and related factorsof Macao residents in China. Information was collected from 983 participants using a structuredquestionnaire with an effective utilization rate of 94.2%. Descriptive statistics, univariate analysisand correlation analysis were used to analyze the data. The results show that, on the one hand,there are significant differences in risk perception on the factors included: (1) age, education and otherdemographic characteristics; (2) health status, occupation, length of stay, residence area, residencefloor, family organization structure and individuals monthly income and other personal or familyconditions; (3) channels and quantity of typhoon information acquisition; (4) degree of mastery ofrelevant risk aversion knowledge. On the other hand, some factors still have a moderate or highlevel of correlation with risk perception: (1) The older the respondent, the lower the education level,the lower the income, the lower the risk perception of property damage, health impact and life threat.(2) The more children or elderly people in the family, the higher the risk perception of respondents.(3) The more risk knowledge, the lower the risk perception. (4) The more channels for obtaininginformation, the lower the fear level and the overall impact of risk perception. (5) The stronger the riskperception, the more positive disaster response behaviors would be taken by the public. In addition,the more information acquisition channels and the less risk knowledge respondents have, the greaterthe risk perception of the overall impact and the fear of the typhoon; the fewer information accesschannels and less risk knowledge respondents have, the greater the risk perceptions of propertydamage, health effects and life threats.Keywords: typhoon disaster; risk perception; knowledge; information acquisition channels;active response to disaster1. IntroductionTropical cyclones are the most destructive disasters in coastal areas together with tsunamis,although the former is more frequent. As a highly destructive weather system, tropical cyclones,whose official warning name is “the wind ball” used in Macao and Hongkong, are cyclonic cyclones thatoccur in tropical or subtropical oceans, which are also often accompanied by strong winds, heavy rainand storm surges. The tropical cyclone warning signals in Hong Kong and Macao are now divided into:Int. J. Environ. Res. Public Health 2020, 17, 7357; doi:10.3390/ijerph17207357 www.mdpi.com/journal/ijerph
Int. J. Environ. Res. Public Health 2020, 17, 7357 2 of 24alert signal No. 1, strong wind signal; No. 3, gale or storm signal; No. 8, gale or storm enhancementsignal, No. 9; and hurricane signal, No. 10. The World Meteorological Organization (WMO) definestropical cyclones with central continuous wind speeds between 32.7 m/s and 41.4 m/s as typhoonsor hurricanes. In recent years, with global warming, typhoon disasters have occurred frequently,which have caused a serious impact on the entire society [1,2]. In accordance with the recommendationsof World Meteorological Organization (WMO), tropical cyclones are classified according to maximumsustained wind near the center of the storm. Based on the 10-min average wind speed, typhoon hasbeen divided into six categories in Macao since 2009, namely, tropical depression, tropical storm,severe tropical storm, typhoon, severe typhoon and super typhoon (Super T.).China is one of the countries most severely affected by typhoon disasters. As a coastal area and alow-altitude city in the southeast corner of China, Macao used to have nearly 10 bays, making it one ofthe most severely affected areas in China (See Figure 1 below for details).Int. J. Environ. Res. Public Health 2020, 17, x 2 of 26 that occur in tropical or subtropical oceans, which are also often accompanied by strong winds, heavy rain and storm surges. The tropical cyclone warning signals in Hong Kong and Macao are now divided into: alert signal No. 1, strong wind signal; No. 3, gale or storm signal; No. 8, gale or storm enhancement signal, No. 9; and hurricane signal, No. 10. The World Meteorological Orga izatio (WMO) defines tropical cyclones with central continuous wind speeds between 32.7 m/s and 41.4 m/s as typhoons or hurricanes. In recent years, with global warming, typhoon disasters have occurred frequently, which have caused a serious impact on the entire society [1,2]. In accordance with the recommendations of World Meteorological Organization (WMO), tropical cyclones are classified according to maximum sustai ed wind near the center of the storm. Base on the 10-min average wind speed, typhoon has been divided into six categories in Macao since 2009, namely, tropical depression, tropical storm, severe tropical storm, typhoo , severe typhoon and super typhoon (Super T.). i a is e f t e c tries st se erely affected by typh on disasters. As a coastal area and a low-altitude city in the southeast corner of China, Macao used to have nearly 10 bays, maki g it one of the most severely affected areas in China (See Figure 1 below for details). Figure 1. Map. (a) Map of China. (b) Map of Macao. In 2017 and 2018, the two consecutive years, Macao experienced three super typhoons—Hato, Mangkhut and Yutu—which caused not only economic losses and casualties, but also some psychological trauma to people. Studies found that some people were more vulnerable because they could not respond reasonably or were not good at adapting, while others were the opposite in the face of the same disaster or environmental change [3]. One of the important reasons is that different people have different perceptions of disasters or environmental changes. These differences could perform significant impact on what kind of mitigation decisions will be made by such differences, what mitigation measures will be taken and what will happen to these measures [3–5]. Researching people’s perception of disasters is the primary basis for a deep understanding of their adaptation to disasters. Slovic believes that when applying risk assessment to evacuate hazards, most people rely on intuitive risk judgment, that is, risk perception [6]. He also summarized risk characteristics into unknown risks and fear risks through factor analysis. Sjöberg believes that risk perception refers to a subjective assessment of the probability of a particular accident and how much we are associated with this adverse outcome [7]. He also believes that cognitive risks include both an assessment of probability and the severity of negative outcomes. Risk perception as a cross-disciplinary and multidisciplinary subject has achieved certain results through decades of research by scholars all over the world. Previous studies have found that public risk perception is related to personal basic conditions, risk knowledge, mass media methods and disaster experience. For example, Lai et al., in a study of Hong Kong, found that women, older 1500 2000 km Fig re 1. Map. (a) Map of China. (b) Map of Macao. In 2017 and 2018, the two consecutiveyears, Macao xp rienced three super typhoons—Hato, Mangkhut and Y tu—which caused not onlyec ic l sses and casualties, but also some psychological trauma to pe ple. Studies found that somepeople were more vulnerable because they could not respond reasonably or were not good at adapting,while others were the opposite in the face of the same disaster or environmental change [3]. One ofthe important reasons is that different people have different perceptions of disasters or environmentalchanges. These differences could perform significant impact on what kind of mitigation decisionswill be made by such differences, what mitigation measures will be taken and what will happen tothese measures [3–5]. Researching people’s perception of disasters is the primary basis for a deepunderstanding of their adaptation to disasters.Slovic believes that when applying risk assessment to evacuate hazards, most people rely onintuitive risk judgment, that is, risk perception [6]. He also summarized risk characteristics intounknown risks and fear risks through factor analysis. Sjöberg believes that risk perception refers toa subjective assessment of the probability of a particular accident and how much we are associatedwith this adverse outcome [7]. He also believes that cognitive risks include both an assessment ofprobability and the severity of negative outcomes.Risk perception as a cross-disciplinary and multidisciplinary subject has achieved certain resultsthrough decades of research by scholars all over the world. Previous studies have found that public riskperception is related to personal basic conditions, risk knowledge, mass media methods and disasterexperience. For example, Lai et al., in a study of Hong Kong, found that women, older respondents andless educated respondents perceived greater environmental risks [8]. Furthermore, researchers foundthat income has a significant positive effect on risk perception, that is, the higher the income is,
Int. J. Environ. Res. Public Health 2020, 17, 7357 3 of 24the smaller the risk perception will be [9]. Jackson studied earthquake disasters on the west coastof North America and found that the public’s ability to understand disasters is closely related todisaster experience and knowledge of disaster events [10]. Bettman and Park showed that the morecomprehensive an individual’s knowledge of risk is, the more it can reduce the uncertainty of risk andthe lower its risk perception [11]. Walter et al. found in researching typhoon disasters that “the morepeople experience typhoon disasters, the higher their perception of risk” [12]. Wei et al. proposedthat the number of media reports affect the public’s risk perception based on the theory of publicmemory [13].In the field of risk management, some scholars found that there is a significant positive correlationbetween risk perception and active disaster response behavior [14–16]. When studying the copingbehavior of the public during typhoons, Riad, Norris and Ruback found that the public’s decisionnot to evacuate the danger zone is affected by the psychological processes of risk perception andsocial influence [14]. Matyas, Srinivasa and Cahyanto researched the risk perception and evacuationdecisions of tourists in hurricane-affected areas, and found that the higher the risk perception is,the stronger their willingness to evacuate is, and coastal tourists have a higher willingness to evacuatethan the mainland tourists [15]. However, some scholars have found that high risk perception does notnecessarily result in defensive behaviors in response to natural disasters [17]. Therefore, understandingthe public’s perception of disaster risk help us understand why and how they use their own methodsto deal with disasters [18].From the current research status, although in-depth research on risk perception has providedscholars with a wealth of policy recommendations for risk management, the authors found thatthere is relatively little literature on the risk perception research of typhoon-type meteorologicaldisasters. The predictability makes the public controllable to a certain extent in response to typhoondisasters, which is different from other sudden natural disasters. Under the circumstances of the publichaving sufficient time to prepare for the risk, their perception of risk and behavioral tendencies needfurther research.Therefore, the purpose of this study is to reveal the risk perception and its influencing factors oftyphoon disasters among residents of typhoon-stricken areas (Macao) through quantitative research,and to further explore the relationship between public risk perception and enthusiasm for disasterresponse. Specifically, it analyzed the risk perception and related factors of Macao residents in China,such as demographic characteristics, personal or family conditions, risk knowledge and disasterresponse behaviors, by testing the following assumptions.Hypothesis 1 (H1). Participants with different demographic characteristics are of significant differences inrisk perception.Hypothesis 2 (H2). Participants in different personal situations are of significant differences in risk perception.Hypothesis 3 (H3). Participants with different levels of risk knowledge are of significant differences inrisk perception.Hypothesis 4 (H4). The number of typhoon information acquisition channels significantly positively affectsparticipants’ risk perception.Hypothesis 5 (H5). Participants’ risk perception significantly positively affects their motivation to respondto disasters.
Int. J. Environ. Res. Public Health 2020, 17, 7357 4 of 242. Materials and Methods2.1. Study SettingMacao covers an area of 32.9 square kilometers. By the end of 2019, the population was 679,600,with a population density of 20,657 per square kilometer. “Parish” is its administrative division unit.There are currently seven parishes and one non-parish division area, including Our Lady FatimaParish, St. Anthony Parish, Sé Freguesias, St. Lazarus Parish, St. Lawrence Parish, Ilha da Taipa,Concelho das Ilhas and Cotai. The lowest point in the region is the sea level of the South China Sea,with an altitude of 0 m. The highest point is Koloane Alto, 172.4 m above sea level. According to theMacao Private Housing Activity Center, the three parishes of Our Lady Fatima Parish, St. LawrenceParish and Concelho das Ilhas are vulnerable areas to typhoons.Macao has a subtropical monsoon climate, but also has the characteristics of a tropical climate.Spring and summer are humid and rainy, and autumn and winter have low relative humidity andless rainfall. The typhoon season is from May to October, with the most frequent period from July toSeptember. From the number of typhoon landings in Macao from 1954 to 2019, 190 typhoons landed inMacao (only typhoons, severe typhoons, and super typhoons with a maximum average wind of morethan 118 km per hour) in 65 years, with an average of 2.92 landings each year. Tropical depressions,tropical storms and severe tropical storms are not included here. Moreover during 2010 to 2019, therewere 18 severe typhoons and super typhoons, with an average of twice a year. See Figure 2 and Table 1for more details.Int. J. Environ. Res. Public Health 2020, 17, x 4 of 26 Macao covers an area of 32.9 square kilometers. By the end of 2019, the population was 679,600, with a population density of 20,657 per square kilometer. “Parish” is its administrative division unit. There are currently seven parishes and one non-parish division area, including Our Lady Fatima Parish, St. Anthony Parish, Sé Freguesias, St. Lazarus Parish, St. Lawrence Parish, Ilha da Taipa, Concelho das Ilhas and Cotai. The lowest point in the region is the sea level of the South China Sea, with an altitude of 0 m. The highest point is Koloane Alto, 172.4 m above sea level. According to the Macao Private Housing Activity Center, the three parishes of Our Lady Fatima Parish, St. Lawrence Parish and Concelho das Ilhas are vulnerable areas to typhoons. Macao has a subtropical monsoon climate, but also has the characteristics of a tropical cli ate. Spring and summer are humid and rainy, and autumn and winter have low relative humidity and less rainfall. The typhoon season is from May to October, with the most frequent period from July to September. From the number of typhoon landings in Macao from 1954 to 2019, 190 typhoons landed in Macao (only typhoons, severe typhoons, and super typhoons with a maximum average wind of more than 118 km per hour) in 65 years, with an average of 2.92 landings each year. Tropical depressions, tropical storms and severe tropical storms are not included here. Moreover during 2010 to 2019, there were 18 severe typhoons and super typhoons, with an average of twice a year. See Figure 2 and Table 1 for more details. Figure 2. Number of typhoon landings (1954–2019). Table 1. Statistics of Severe Typhoons and super typhoons (2010–2020). Type Name Month Severe Typhoon Fanabi Sep. 2010 Super Typhoon Megi Oct. 2010 Severe Typhoon Nigg Oct. 2011 Severe Typhoon Vicente July. 2012 Severe Typhoon Tembin Aug. 2012 Super Typhoon Utor Aug. 2013 Super Typhoon Usagi Sep. 2013 Severe Typhoon Krosa Nov. 2013 Super Typhoon Rammasun July. 2014 Severe Typhoon Mujigea Oct. 2015 Super Typhoon Meranti Sep. 2016 Severe Typhoon Megi Sep. 2016 Super Typhoon Sarika Oct. 2016 Super Typhoon Haima Oct. 2016 Super Typhoon Hato Aug. 2017 Severe Typhoon Khanun Oct. 2017 Super Typhoon Mangkhut Sep. 2018 Super Typhoon Yutu Oct. 2018 0123456789195419561958196019621964196619681970197219741976197819801982198419861988199019921994199619982000200220042006200820102012201420162018FrequencyYearFigure 2. u ber of typhoon landings (1954–2019).On 16 September 20 8, the strongest typhoon “Ma gkhut” wreaked havoc in Macao for more than24 h, with typhoon No. 10 taking effect for up to 9 h, bringing storms and floods to Macao, especially thelow-lying areas, such as the inner harbor and the ocean. Shops were flooded, the infrastructure weredamaged and trees fell heavily. “Mangkhut” brought about an estimated direct economic loss of520 million MOP (65 million USD) and an i direct loss up to 1.03 billion MOP (130 million USD),i.e., a total loss of 1.55 billion MOP (200 milli USD). Th Macao Civil Defense Center received a totalof 445 accident reports and at least 18 people were injured, and one year before, Macao was hit by thestrongest typhoon “Hato” since the typhoon observations were made in 1953. (1) Hato was a little bitweaker than Mangkhut, but caused a much more serious damage; (2) Electricity and communicationswere paralyzed in a large area, and property damage reached 11.47 billion MOP (1.44 billion USD);(3) Most important of all, 10 people were killed and 244 were injured, which was very serious for sucha small city as Macao.
Int. J. Environ. Res. Public Health 2020, 17, 7357 5 of 24Table 1. Statistics of Severe Typhoons and super typhoons (2010–2020).Type Name MonthSevere Typhoon Fanabi September 2010Super Typhoon Megi October 2010Severe Typhoon Nigg October 2011Severe Typhoon Vicente July 2012Severe Typhoon Tembin August 2012Super Typhoon Utor August 2013Super Typhoon Usagi September 2013Severe Typhoon Krosa November 2013Super Typhoon Rammasun July 2014Severe Typhoon Mujigea October 2015Super Typhoon Meranti September 2016Severe Typhoon Megi September 2016Super Typhoon Sarika October 2016Super Typhoon Haima October 2016Super Typhoon Hato August 2017Severe Typhoon Khanun October 2017Super Typhoon Mangkhut September 2018Super Typhoon Yutu October 2018Severe Typhoon Wutip February 2019Severe Typhoon Lekima August 2019Severe Typhoon Lingling September 2019Severe Typhoon Hagibis October 2019Severe Typhoon Bualoi October 2019Super Typhoon Halong November 2019Severe Typhoon Kammuri December 20192.2. Study ParticipantsThe study population needed to have lived in Macao for at least six months, have typhoonexperience, either male or female, and have clear consciousness, no mental disorders and the ability toexpress their views properly. After the investigators’ explanation, they can understand the purpose ofour survey, read each question and understand the meaning of their corresponding options and theycan make their own choices; or if they cannot read, they can ask investigators to fill it out for them withtheir own choices after the investigators’ explanations.A total of 1100 questionnaires were distributed in this study, and 1043 were recovered with arecovery rate of 94.8%. Among them, a total of 983 participants had experienced the typhoons of“Mangkhut” or “Hato”, which is an effective rate of 94.2%.2.3. Data Collection2.3.1. Survey DesignA structured questionnaire was compiled after consulting relevant literature and conductingpreliminary interviews with typhoon experts and local residents. The questionnaire was designedfrom the following aspects:(1) Screening items, whether they have experienced typhoon “Mangkhut” or “Hato”;(2) Demographic characteristics, such as gender, age, education, type of identity, occupation, etc.;(3) Other personal or family conditions, such as physical health, living area, living floor, length ofstay, family structure, personal monthly income, monthly family income, etc.;(4) Channels to obtain typhoon information, including TV, SMS (Short Message Service), Internet,radio, posting information, newspapers, friends and relatives, loud speakers, etc.;(5) Risk knowledge, such as understanding of signal lights, knowledge on typhoon and how toprevent possible risks caused by typhoon, and understanding of disaster response behaviors;
Int. J. Environ. Res. Public Health 2020, 17, 7357 6 of 24(6) Risk perception, examining the participants’ five perceptions of risk before thetyphoon—the overall impact, property damage, health effect, life threat, and fear level;(7) What disaster response actions have been taken?The five-point Likert scale was used for the two categories of degree judgment and attitudeexpression. For example, in terms of residents’ awareness of the impact of typhoon disasters, the fiveitems of “overall impact,” “property damage”, “life threat”, “health effect” and “degree of fear”, etc.“Very large/Very strong”, “relatively large/relatively strong”, “average”, “relatively small/relativelyweak” to “very small/very weak”, in order of “5, 4, 3, 2, 1”.2.3.2. Pre-InvestigationIn May 2019, members of the research team randomly selected 56 Macao residents who hadexperienced typhoon disaster to conduct a preliminary survey. The combined reliability of the riskperception part of the questionnaire was 0.9573, and the average variance extracted (AVE) was 0.7907,which had good reliability and validity. Finally, the questionnaire was further modified and improvedaccording to the feedback from the pre-survey. After review by relevant experts, it was revised againto form the final questionnaire.2.4. Statistical AnalysisStatistical analysis was performed using Statistical Product and Service Solutions (SPSS, IBM,New York, NY, USA).(1) Descriptive statistics was used to describe participants’ demographic characteristics,other personal or family conditions, channels for obtaining typhoon information, risk knowledge,and disaster response behaviors.(2) Means and standard deviations were used to describe participants’ risk perception scores.(3) T-test or F-test were used to test whether there are significant differences in risk perceptionamong participants with different demographic characteristics, different personal or family conditions,different levels of knowledge of risk knowledge, and different numbers of typhoon informationacquisition channels.(4) Pearson correlation analysis was used to test whether the five items of risk perception weresignificantly correlated. Pearson correlation analysis was used to detect whether typhoon knowledgeaffected risk perception and whether the number of typhoon information acquisition channels affectedrisk perception.(5) Variance analysis was used to test whether participants with different risk perceptions hadsignificantly different enthusiasm for disaster response.3. Results and DiscussionIn this section, the results are presented and discussed in relation to (1) the distribution ofparticipants’ demographic characteristics and other personal and family conditions; (2) the typhooninformation section, which contains the description statistics of the participants’ typhoon disasterprevention knowledge and typhoon information acquisition channels; (3) descriptive statistics ofrespondents’ risk perception before the typhoon; (4) descriptive statistics of respondents’ active responseto typhoon; and (5) multivariable analysis among factors influencing residents’ risk perception.3.1. Demographic Characteristics and Other Information of Participants3.1.1. Demographic Characteristics of ParticipantsFigure 3 shows the demographic characteristics of participants. The proportions of men andwomen in the participants were basically the same, with more people in the 25–30 and 65-year-oldgroups. Nearly half of the participants with a college degree in the valid questionnaire recovered.
Int. J. Environ. Res. Public Health 2020, 17, 7357 7 of 24In terms of occupational distribution, there are more professionals (19.8%), students (17%) andunemployed persons (19.6%) than other occupations.Int. J. Environ. Res. Public Health 2020, 17, x 7 of 26 (a) (b) (c) (d) Figure 3. Demographic characteristics of the participants (n = 983). (a) Gender distribution of participants. (b) Age distribution of participants. (c) Distribution of participants’ education levels. (d) distribution of participants’ occupation. 3.1.2. Other Personal Information of Participants Figure 4 shows that more than 80% of the participants consider their health condition relatively good (36.1%) or very good (47.2%). There are a total of eight parishes in Macao: Our Lady Fatima Parish, St. Anthony Parish, Sé Freguesias, St. Lazarus Parish, St. Lawrence Parish, Ilha da Taipa, Concelho das Ilhas and Cotai. A total of 78.2% of the participants lived in Macao and most of the participants in Macao lived in the following three parishes: Our Lady Fatima Parish (20.2%), St. Anthony Parish (16.5%) and Sé Freguesias (19.5%). When it comes to living floors, nearly half of the participants said they lived on the second to seventh floors. Due to our initial consideration of the Figure 3. Demographic characteristics of the participants (n = 983). (a) Gender distribution ofparticipants. (b) Age distribution of participants. (c) Distribution of participants’ education levels.(d) distribution of participants’ occupation.3.1.2. Other Personal Information of ParticipantsFigure 4 shows that more than 80% of the participants consider their health condition relativelygood (36.1%) or very good (47.2%). There are a total of eight parishes in Macao: Our Lady Fatima Parish,St. Anthony Parish, Sé Freguesias, St. Lazarus Parish, St. Lawrence Parish, Ilha da Taipa, Concelho dasIlhas and Cotai. A total of 78.2% of the participants lived in Macao and most of the participants inMacao lived in the following three parishes: Our Lady Fatima Parish (20.2%), St. Anthony Parish
Int. J. Environ. Res. Public Health 2020, 17, 7357 8 of 24(16.5%) and Sé Freguesias (19.5%). When it comes to living floors, nearly half of the participants saidthey lived on the second to seventh floors. Due to our initial consideration of the structure of thequestionnaire, most of the respondents stated that they were native Macao people (34.4%) or hadbeen living in Macao for more than ten years (28.6%). When it comes to family structure, the majorityof respondents said there was no family member with the age of under 15 years of age (46.7%) orover 65 years old (54.0%), and 80.4% of respondents said they had no family members with mobilityproblems. In terms of economic status, 40% of the respondents had a personal monthly income below5000 MOP (~600 USD), 29.9% of the respondents had a personal monthly income between 5000 MOP(~600 USD) and 15,000 MOP (~1800 USD).3.2. Typhoon Information3.2.1. Knowledge of Typhoon Disaster PreventionMany scholars have shown in their research that improving residents’ knowledge of risk aversionhelps to reduce the potential risks of typhoons [19]. It is officially stipulated that the suspension oftyphoon signal (i.e., wind ball) No. 8 and above requires advance prevention and paid holidays. As isshown in Figure 5, only 47% of the respondents believe that a typhoon with a typhoon signal No. 8 andabove need to be prepared in advance. Although more than half of the respondents indicated that theyknew well (37%) or they were very familiar with (18.4%) precautions, 21.8% of the respondents statedthat “when typhoon comes, rescuing property already in dangerous areas” can reduce losses, which isactually a risky activity. Most respondents believe that measures to reduce losses are pre-disasterpreparations, such as consolidating doors, windows and outdoor facilities (88.6%), turning off electricalappliances (65.8%), preparing emergency supplies, such as food and medicine (76.8%), parking theirown vehicle in a secure parking lot in advance (74.2%) and staying away from dangerous and low-lyingareas (74.4%).3.2.2. Information Channels on Risk CommunicationFigure 6 shows that TV (82.3%) and the internet (78.2%) were the main channels for most residentsto obtain typhoon information before the typhoon arrives. Furthermore, the respondents also receivedthe information through SMS (54.0%), radio (38.7%) or friends or relatives (49.4%). This result basicallycoincides with AlQahtany’s results, which revealed that television, mobile phone text message and theinternet rank among the top three existing methods of obtaining information about disaster risks [20].Compared with other research, in addition to traditional media–television, the internet has become avery important channel for people to obtain typhoon information [13,21–25].3.3. Risk Perception of Residents before TyphoonFigure 7 shows respondents’ risk perception of typhoon before its landfall with regard to overallimpact, property damage, health effect, life threat and fear level. A total of 81.4% of them thought theoverall impact of typhoon was big or very big (scored 4 or 5 out of maximum 5). About one-third ofrespondents had a high-risk perception of the risks associated with property damage, health effects,and life threats, while, 64.9% of them indicated that their fear level was high or very high. From anaverage point of view, the overall impact and fear level of respondents on risk perception, which wasaround 4, was higher than other three dimensions, including property damage, life threatening andhealth effects, with scores ranging from 2.5 to 3.
Int. J. Environ. Res. Public Health 2020, 17, 7357 9 of 24Int. J. Environ. Res. Public Health 2020, 17, x 8 of 26 structure of the questionnaire, most of the respondents stated that they were native Macao people (34.4%) or had been living in Macao for more than ten years (28.6%). When it comes to family structure, the majority of respondents said there was no family member with the age of under 15 years of age (46.7%) or over 65 years old (54.0%), and 80.4% of respondents said they had no family members with mobility problems. In terms of economic status, 40% of the respondents had a personal monthly income below 5000 MOP (~600 USD), 29.9% of the respondents had a personal monthly income between 5000 MOP (~600 USD) and 15,000 MOP (~1800 USD). (a) (b) (c) a (d) (e) Int. J. Environ. Res. Public Health 2020, 17, x 9 of 26 (f) (g) (h) (i) b Figure 4. Other personal information of participants (n = 983). (a) Distribution of participants’ health condition. (b) Distribution of participants’ place of residence. (c) Distribution of participants’ residential zone. (d) Distribution of participants’ living floors. (e) Distribution of participants’ length of stay. (f) Distribution of the number of family members under the age of 14 of the participants. (g) distribution of participants. (h) distribution of participants. (i) distribution of participants. a If the place of residence is outside Macao, skip this question; b missing system 16. 3.2. Typhoon Information 3.2.1. Knowledge of Typhoon Disaster Prevention Many scholars have shown in their research that improving residents’ knowledge of risk aversion helps to reduce the potential risks of typhoons [19]. It is officially stipulated that the suspension of typhoon signal (i.e., wind ball) No. 8 and above requires advance prevention and paid holidays. As is shown in Figure 5, only 47% of the respondents believe that a typhoon with a typhoon signal No. 8 and above need to be prepared in advance. Although more than half of the respondents indicated that they knew well (37%) or they were very familiar with (18.4%) precautions, 21.8% of the respondents stated that “when typhoon comes, rescuing property already in dangerous areas” can reduce losses, which is actually a risky activity. Most respondents believe that measures to reduce losses are pre-disaster preparations, such as consolidating doors, windows and outdoor facilities (88.6%), turning off electrical appliances (65.8%), preparing emergency supplies, such as food and medicine (76.8%), parking their own vehicle in a secure parking lot in advance (74.2%) and staying away from dangerous and low-lying areas (74.4%). Figure 4. Other personal information of participants (n = 983). (a) Distribution of participants’ healthcondition. (b) Distribution of participants’ lace of r sidence. (c) Distributio of p rticipants’ residentialzone. (d) Distribution of participants’ living floors. (e) Distribution of participants’ length of stay.(f) Distribution of the number of family members under the age of 14 of the participants. (g) distributionof participants. (h) distributi n of participan s. (i) d stribution of participants. a If the place of residenceis outside Macao, skip this question; b missing system 16.
Int. J. Environ. Res. Public Health 2020, 17, 7357 10 of 24Int. J. Environ. Res. Public Health 2020, 17, x 10 of 26 (a) (b) (c) a Figure 5. Participants’ knowledge of typhoon disaster precautions (n = 983). (a) Distribution of “When the typhoon signal is _ or above, one should get prepared for typhoon in advance.” (b) Distribution of “Do you know how to guard against typhoons?” (c) Distribution of “Which of the following measures do you think will reduce your loss?” a Percentage total may add up to more than 100% as multiple responses were permissible. 3.2.2. Information Channels on Risk Communication Figure 6 shows that TV (82.3%) and the internet (78.2%) were the main channels for most residents to obtain typhoon information before the typhoon arrives. Furthermore, the respondents also received the information through SMS (54.0%), radio (38.7%) or friends or relatives (49.4%). This result basically coincides with AlQahtany’s results, which revealed that television, mobile phone text message and the internet rank among the top three existing methods of obtaining information about disaster risks [20]. Compared with other research, in addition to traditional media–television, the internet has become a very important channel for people to obtain typhoon information [13,21–25]. Figure 5. Participants’ knowledge of typhoon disaster precautions (n = 983). (a) Distribution of “Whenthe typhoon signal is _ or above, one should get prepared for typhoon in advance.” (b) Distribution of“Do you know how to guard against typhoons?” (c) Distribution of “Which of the following measuresdo you think will reduce your loss?” a Percentage total may add up to more than 100% as multipleresponses were permissible.Int. J. Environ. Res. Public Health 2020, 17, x 11 of 26 Figure 6. Participants’ information channels for typhoon (n = 983) a. a Percentage total may add up to more than 100% as multiple responses were permissible. 3.3. Risk Perception of Residents before Typhoon Figure 7 shows respondents’ risk perception of typhoon before its landfall with regard to overall impact, property damage, health effect, life threat and fear level. A total of 81.4% of them thought the overall impact of typhoon was big or very big (scored 4 or 5 out of maximum 5). About one-third of respondents had a high-risk perception of the risks associated with property damage, health effects, and life threats, while, 64.9% of them indicated that their fear level was high or very high. From an average point of view, the overall impact and fear level of respondents on risk perception, which was around 4, was higher than other three dimensions, including property damage, life threatening and health effects, with scores ranging from 2.5 to 3. (a) (b) (c) (d) Figure 6. Participants’ information channels for typhoon (n = 983) a. a Percentage total may add up tomore than 100% as multiple responses were permissible.
Int. J. Environ. Res. Public Health 2020, 17, 7357 11 of 24Int. J. Environ. Res. Public Health 2020, 17, x 11 of 26 Figure 6. Participants’ information channels for typhoon (n = 983) a. a Percentage total may add up to more than 100% as multiple responses were permissible. 3.3. Risk Perception of Residents before Typhoon Figure 7 shows respondents’ risk perception of typhoon before its landfall with regard to overall impact, property damage, health effect, life threat and fear level. A total of 81.4% of them thought the overall impact of typhoon was big or very big (scored 4 or 5 out of maximum 5). About one-third of respondents had a high-risk perception of the risks associated with property damage, health effects, and life threats, while, 64.9% of them indicated that their fear level was high or very high. From an average point of view, the overall impact and fear level of respondents on risk perception, which was around 4, was higher than other three dimensions, including property damage, life threatening and health effects, with scores ranging from 2.5 to 3. (a) (b) (c) (d) Int. J. Environ. Res. Public Health 2020, 17, x 12 of 26 (e) (f) Figure 7. Knowledge, information channels and active response to typhoon and risk perception. (a) Distribution of participants’ expected overall impact of typhoon. (b) Distribution of participants’ expected property damage of typhoon. (c) Distribution of participants’ expected harm on individual health. (d) Distribution of participants’ expected threaten on individual life. (e) Distribution of participants’ fear of typhoon. (f) The mean (M) and standard deviation (SD) of different dimension of risk perception. 3.4. Active Response for Typhoon As is shown in Figure 8, the top five measures for active disaster response are: consolidation of doors, windows and outdoor facilities (94.4%), preparation for emergency supplies such as food and medicine (93.4%), avoidance of outside activity (89.2%), protection of personal wealth and vehicles ahead of time (88.7%), staying away from dangerous and low-lying areas (87.1%). In the previous study of Typhoon Mangkhut by Chan et al., similar conclusions were reached [26]. However, the number of people who choose to go out to a safe haven is the least, only 7.4%. This may be due to the government’s stipulation that if the hoisting of No. 8 typhoon signal is announced during work hours, workers should leave work in batches, and the government will give a notice two hours in advance to facilitate smooth traffic. If it is suspended before work in the morning, companies and schools will automatically suspend work and classes, and the Hong Kong stock market will also be suspended. When Typhoon No. 8 turns into a weaker Typhoon No. 3, employees need to return to the company within two hours. Figure 8. Participant’s active response to typhoon (n = 983) a. a Percentage total may add up to more than 100% as multiple responses were permissible. 3.5. Factors Influencing Residents’ Risk Perception—Multivariable Analysis Figure 7. Knowledge, information channels and active response to typhoon and risk perception.(a) Distribution of participants’ expected overall impact of typhoon. (b) Distribution of participants’expected property damage of typhoon. (c) Distribution of participants’ expected harm on individualhealt . (d) Distribution of participants’ ex t t reaten on individual life. (e) Distribution ofparticipants’ fear of typhoon. (f) The mean ( ) standard deviation (SD) o different dimension ofrisk perception.3.4. Active Response for TyphoonAs is shown in Figure 8, the top five measur f ctive disaster response are: c nsolidation ofdoors, windows and outdoor facilities (94.4%), preparation for emergency supplies such as food andmedicine (93.4%), avoidance of outside activity (89.2%), protection of personal wealth and vehiclesahead of time (88.7%), staying away from dangerous and low-lying areas (87.1%). In the previous studyof Typhoon Mangkhut by Chan et al., similar conclusions were reached [26]. However, the numberof people who choose to go out to a safe haven is the least, only 7.4%. This may be due to thegovernment’s stipulation that if the hoisting of No. 8 typhoon signal is announced during work hours,workers should leave work in batche , a the g vernme t will give a n tice two hours in advance tofacilitate smooth traffic. If it is suspend d before w rk in the morning, comp nies and schools willautomatically suspend work a d classes, and the Hong Kong stock market will also be suspended.When Typhoon No. 8 turns into a weaker Typhoon No. 3, employees need to return to the companywithin two hours.
Int. J. Environ. Res. Public Health 2020, 17, 7357 12 of 24Int. J. Environ. Res. Public Health 2020, 17, x 12 of 26 (e) (f) Figure 7. Knowledge, information channels and active response to typhoon and risk perception. (a) Distribution of participants’ expected overall impact of typhoon. (b) Distribution of participants’ expected property damage of typhoon. (c) Distribution of participants’ expected harm on individual health. (d) Distribution of participants’ expected threaten on individual life. (e) Distribution of participants’ fear of typhoon. (f) The mean (M) and standard deviation (SD) of different dimension of risk perception. 3.4. Active Response for Typhoon As is shown in Figure 8, the top five measures for active disaster response are: consolidation of doors, windows and outdoor facilities (94.4%), preparation for emergency supplies such as food and medicine (93.4%), avoidance of outside activity (89.2%), protection of personal wealth and vehicles ahead of time (88.7%), staying away from dangerous and low-lying areas (87.1%). In the previous study of Typhoon Mangkhut by Chan et al., similar conclusions were reached [26]. However, the number of people who choose to go out to a safe haven is the least, only 7.4%. This may be due to the government’s stipulation that if the hoisting of No. 8 typhoon signal is announced during work hours, workers should leave work in batches, and the government will give a notice two hours in advance to facilitate smooth traffic. If it is suspended before work in the morning, companies and schools will automatically suspend work and classes, and the Hong Kong stock market will also be suspended. When Typhoon No. 8 turns into a weaker Typhoon No. 3, employees need to return to the company within two hours. Figure 8. Participant’s active response to typhoon (n = 983) a. a Percentage total may add up to more than 100% as multiple responses were permissible. 3.5. Factors Influencing Residents’ Risk Perception—Multivariable Analysis Figure 8. Participant’s active response to typhoon (n = 983) a. a Percentage total may add up to morethan 100% as multiple responses were permissible.3.5. Factors Influencing Residents’ Risk Perception—Multivariable AnalysisPearson correlation analysis was used to test five topics related to risk perception, and Table 2below shows that a significant positive correlation between each of them at p < 0.001. Among them,the degree of fear and property damage have a higher impact on the risk perception of the overallimpact of the respondents, with correlation coefficients of 0.369 and 0.330, respectively. Moreover,there is a strongly significantly positive correlation between the respondents’ perceptions of the risk ofproperty damage and their risks to health and life threats, with correlation coefficients of 0.777 and0.701, respectively. The correlation coefficient for health effect and life threat risk perception is also ashigh as 0.822. First of all, health and life are closely related to each other. It is generally believed thatthreats to life generally endanger health, so the two have a strong positive correlation. Furthermore,once the typhoon has an impact on health or life, the individual’s property will naturally suffer from acorresponding loss.Table 2. Correlation between items of risk perception.Overall Impact Property Damage Health Effects Life Threat Fear LevelOverall impact Pearson 1 - - - -Property damage Pearson 0.330 *** 1 - - -Health effects Pearson 0.251 *** 0.777 *** 1 - -Life threat Pearson 0.207 *** 0.701 *** 0.822 *** 1 -Fear level Pearson 0.369 *** 0.298 *** 0.291 *** 0.316 *** 1*** p < 0.001.3.5.1. Comparison Analysis of Risk Perception based on Demographic Characteristics and CorrelationAnalysis between Them(1) Risk perception about overall impactAs is shown in Table 3, respondents of different ages had significant differences in their perceptionof the overall impact of the typhoon before typhoon arrives at p < 0.05, while respondents of differentoccupations had significant differences in their perception of the overall impact of the typhoon beforethe onset of the typhoon at p < 0.01. Among them, middle-aged respondents (35–54 years old) hada higher average risk perception of the overall impact of typhoons, which was about 4.3. When theindividual is 35 to 59 years old, that is, after entering middle age, the physical and behavioralmaladjustment and psychological imbalance will occur, which is what we often call the “middleage crisis”. Middle-aged people are generally the backbone of the family and have heavier familyresponsibilities, because they need to not only raise their children, but also support their parents.This may cause respondents in this age group to pay more attention to the overall impact of typhoons,so their average risk perception score is higher. As far as occupations are concerned, only the risk
Int. J. Environ. Res. Public Health 2020, 17, 7357 13 of 24perception of the overall impact of typhoons of service and sales and students are less than four points,which may be because such respondents who are mainly immigrant labors or students from MainlandChina have not stayed at Macao for a long time. They do not have a sense of regional belonging,so they were not very concerned about the impact of the typhoon. Or, maybe because most of them arefrom inland and have not experienced typhoon before.(2) Risk perception about property damageAs is shown in Table 3, respondents of different genders, ages, education and occupations havesignificant differences in their perception of risk to property damage, at p < 0.001. In terms of gender,men’s risk perception scores for property damage are significantly higher than women’s, which may bebecause men bear more household expenses; this result is contrary to Lai’s research on Hong Kong [8].The 25–44 age group perceived a greater risk of property damage than other age groups, with the agegroup over 65 years the least aware of it. From the perspective of education, respondents with a highersecondary education have a higher risk perception of property damage and the unemployed have thelowest risk perception score of property damage, which may be because they do not typically own asmuch property. This is basically consistent with previous research results [27].(3) Risk perception about health effectsAs is shown in Table 3, respondents of different genders, ages, education and occupations havesignificant differences in their perception of risk to health effects, at p < 0.001. The differences betweenthe subgroups are basically consistent with the dimension of property damage, and will not berepeated here.(4) Risk perception about life threatAs is shown in Table 3, respondents of different genders, ages, education and occupations havesignificant differences in their perception of risk to life threat, at p < 0.001. The differences between thesubgroups are basically consistent with the dimensions of property damage and health effect, and willnot be repeated here.(5) Risk perception about fear levelAs is shown in Table 3, respondents of different genders, ages, education and occupations havesignificant differences in their perception of risk to fear level, at p < 0.001. Different from the results inthe previous dimensions, respondents over the age of 65, those with primary education and belowhave a significantly higher level of fear of typhoon risk than other groups. This may be due to thelow level of education of this group, the lack of knowledge of risks, the inexplicable fear of naturaldisasters and the uncertainty in how to reduce the risks and the extent of losses.In summary, Hypothesis 1, that “participants with different demographic characteristics are ofsignificant differences in risk perception” is accepted.(6) Correlation between demographic characteristics and risk perceptionAs is shown in Table 3, age and education are related to risk perception. Regarding age, on theone hand, there is a significant and moderate negative correlation between age and the risk perceptionof property loss, health impact and life threat with the correlation coefficients of −0.343, –0.442 and−0.469, respectively. On the other hand, there is a significant low-degree positive correlation betweenage and the degree of fear, the risk perception, with a correlation coefficient of 0.147. In other words,to a certain extent, the older the respondent, the lower the risk perception of the respondents in termsof property damage, health impact and life threats. This conclusion is contrary to the findings ofLai et al. on Hong Kong, which may be due to Macao’s unique social welfare system [8]. Macau’s socialwelfare system can guarantee the livelihoods of its residents: for instance, in 2018, each elderlyperson can receive 69,850 MOP (8728 USD) as pension or cash sharing from the government, which isenough for their daily use. Furthermore, the Macao government pays much attention to the residents’livelihood. Therefore, the elderly residents know that, even if the typhoon has an impact on their lives,the government will provide subsidies in time, so their risk perception is relatively low.
Int. J. Environ. Res. Public Health 2020, 17, 7357 15 of 24Regarding education, on the one hand, there is a significant and moderately positive correlationbetween education and the risk perception of property loss, health impact and life threat, and thecorrelation coefficients are 0.325, 0.413 and 0.421, respectively. On the other hand, there is a significantlow-degree negative correlation between education and the degree of fear in terms of risk perception,with a correlation coefficient of −0.165. Most of the locals or the elderly in Macao have relativelylow academic qualifications, so this result can also be partly attributed to Macao’s unique socialwelfare system.3.5.2. Comparison Analysis of Risk Perception based on other Personal or Family Circumstances andCorrelation Analysis between Them(1) Risk perception about overall impactAs can be seen from Table 4 that respondents with different health conditions have significantlydifferent risk perceptions of the overall impact of the typhoon, at 1% level. Respondents living inhazardous areas have a significantly higher level of risk perception of the overall impact. The presenceof elderly people over 65 or family members with limited mobility at home also affect respondents’perception of the risk of overall impact, at p < 0.05. Respondents who live with family members withlimited mobility have a higher level of risk perception of the overall impact of the typhoon. This is alsovery easy to understand. After all, if there are elderly people or members with limited mobility in thefamily, their evacuation behavior in case of disaster will definitely be affected.(2) Risk perception about property damage, health effects and life threatDue to the high correlation between these three dimensions of property damage, health effectsand life threats, there is also some consistency in the analysis of them and other variables, and they areanalyzed together to prevent redundancy.All the personal and family conditions, including health condition, place of residence, whether tolive in areas susceptible to typhoon, living floor, length of stay, monthly personal income, monthly familyincome and whether there are children under 14 years old, elderly over 65 years old or family memberswith limited mobility affected the respondents’ risk perception of property damage, health effects andlife threat, at 5%, 1% or 0.1% level.First of all, respondents with poor health feel greater threats to property, health and life, which maybe due to the fact that these the respondents believe that once they are in danger, their lives will beworse. Secondly, residents living in Macao have the lowest level of risk perception in these threedimensions. Previous studies have also shown that in areas like Macao where typhoon occur morefrequently (almost every year there are several heavy rains and typhoons), residents’ risk perceptionis low, since although the previous incidents also caused some losses, the fatality rate was low andpersonal property losses were low, which caused a sense of control among residents to a certainextent [19,28]. On the one hand, it may be because local residents have become accustomed to typhoons;on the other hand, typhoons have never directly hit Macao in preceding years, which made them thinkthat Macao is a blessed place. Third, respondents who live in areas susceptible to typhoons or live onthe ground floor and below are more aware of the risks of property damage, health and life threats,which is the same as Alqahtany’s study [20]. This is obvious because these areas or floors are morelikely to be flooded. Fourth, respondents who stayed in Macao for five to ten years are more aware ofthese three risks, which may involve Macao policies. The Macao government requires a time span of7 years for both skilled and married immigrants who have just obtained their non-permanent residentsidentity to become permanent residents and enjoy the benefits for permanent residents. Therefore,some of the respondents who have stayed in Macao for five to ten years are about to acquire or have justobtained permanent resident status on the basis that they have accumulated a certain amount of wealth,so the level of perception of these three risks is higher. Fifth, people with the weakest perception ofthese three risks include respondents who do not have children under 14 years of age, elderly peopleover 65 years of age or members with limited mobility. After all, such a family structure needs tobe considered when taking risk avoidance measures or making decisions. Finally, when it comes to
Int. J. Environ. Res. Public Health 2020, 17, 7357 16 of 24personal income, respondents with an income of less than 5000 MOP (600 USD) have the weakestperception of these three risks, probably because they have nothing to lose, or even if they want toprevent losses, there is no financial support. This result is the same as Kellens, who thought that thesepeople may not consider the risk problem because of the risk avoidance cost [29]. Respondents withincomes ranging from 10,000 MOP (1254 USD) to 14,999 MOP (1881 USD) have the strongest riskperception in these three categories, probably because this group of people has just started their careersand does not want to lose anything due to natural disasters, which will affect their future development.(3) Risk perception about fear levelLiving floor, length of stay, monthly family income and whether there are family members withlimited mobility are all factors impacting the respondents’ risk perception of life threat, at 5%, 1% or0.1% level. First of all, respondents living in areas prone to typhoons and on the ground floor and belowhave a higher degree of fear, mainly of heavy rains and floods that follow the typhoon to overwhelmhouses or roads. Second, respondents who had stayed in Macao for less than one year had the highestlevel of fear, probably because they were not familiar with typhoons. Finally, the interviewees withfamily members with limited mobility have the highest degree of fear. The reasons have been statedabove and will not be repeated here.In general, hypothesis two that “participants in different personal situations are of significantdifferences in risk perception” is accepted.(4) Correlation between personal monthly income and risk perceptionFurther analysis on Spearman’s correlation of other personal information and various dimensionsof risk perception found that, first of all, the length of stay is significantly negatively correlated withthe risk perception of the three dimensions of property damage, health effect and life threatening,and the correlation coefficients for these are −0.126, −0.213, and −0.208, respectively. In other words,the longer a resident stays in Macao, the lower the risk perception. However, the length of stay ispositively correlated with the degree of fear. The shorter the respondent stays in Macao, the morefearful they feel.Second, number of family members under 14 or above 65 is significantly positively correlated withrisk perception in the three dimensions of property damage, health effect and life threat, with correlationcoefficients greater than 0.35, which also means that the more family members under 14 or above 65,the greater the risk perception. Moreover, there is a significant but lower positive correlation betweenthe number of family members over 65 years of age and the risk perception of the overall impact.Third, research shows that the higher the income, the stronger the risk perception of propertydamage, health effects and life threats. Although this result is contrary to Spence, we believe thatrespondents with higher incomes cherish their property, health and life more and therefore have ahigher level of risk perception [9].3.5.3. Comparison Analysis of Risk perception based on Knowledge and Correlation Analysisbetween Them(1) Knowledge of typhoon signalRespondents with different degrees of typhoon signal understanding have significant differencesin the two risk perception dimensions of overall impact and degree of fear, at 5% or 0.1% level, while theunderstanding has no significant correlation with risk perception. There are no significant differencesin the three risk-related dimensions of property damage, health effect and life threatening.
Int. J. Environ. Res. Public Health 2020, 17, 7357 18 of 24(2) Understanding of typhoon preventionThe respondents’ understanding on preventive measures is significantly different in all fivedimensions of risk perception. Not only that, but there is a significant negative correlation in the riskperception of the three dimensions of property damage, health effect and life threat; the correlationcoefficients are −0.182, −0.138 and −0.120, respectively. In other words, the more they know about thepreventive measures for typhoon, the lower their risk perception. This conclusion is consistent withBettman, since the more comprehensive knowledge that an individual has, the lower the uncertaintyof the risk and the lower the risk perception [11].(3) Knowledge of typhoon preparednessThis item is basically consistent with the results of the previous project which shows thatrespondents with different knowledge of typhoon preparedness have significant differences in riskperception in five dimensions, at 1% or 0.1% level. In addition to the overall impact and the degreeof fear, the risk perception of the other three dimensions is significantly negatively correlated withknowledge of typhoon preparedness; the correlation coefficients are −0.187, −0.202 and −0.213,respectively, which is slightly higher than understanding of typhoon prevention. Obviously, the deeperyou understand the preventive measures, the lower your perception of risk. After all, if residentsknow the preventive measures for typhoons well, they know how to avoid risks from the aspects ofproperty, health and life, and minimize the risk of typhoons. The corresponding risk perception willnaturally decrease. Moreover, if residents do not know how to prevent typhoons, they may not be ableto accurately estimate the risk of loss, and may easily overestimate the loss.In summary, hypothesis three that “participants with different levels of risk knowledge are ofsignificant differences in risk perception” is accepted.3.5.4. Comparison Analysis of Risk Perception based on Information Channels and CorrelationAnalysis between ThemAs can be seen from Table 5, the number of information acquisition channels has a statisticallypositive impact on the respondents’ risk perception, at 1% or 0.1% level. Moreover, the moreinformation the interviewees have access to before the typhoon, the lower their level of risk perceptionof health effects and life threat. These findings are consistent with the research of Bettman andPark [11]. Meanwhile, these are exactly the opposite in terms of overall impact and fear. There is apositive correlation between the level of fear and the number of channels for obtaining information;the correlation coefficient is 0.246, at 0.1% level. This may be due to the amplification mechanism ofrisk. Slovic proposed that the magnitude of the impact of a risk event depends not only on the natureof the risk event, but also on how the public obtains relevant information and how to perceive andexplain these messages during the risk communication [6]. Furthermore, based on public memorytheory, Wei et al. suggested that the number of media reports affect the public’s perception of risk [13].In summary, hypothesis four that “the number of typhoon information acquisition channelssignificantly positively affects participants’ risk perception” is accepted.3.5.5. Comparison Analysis of Active Response to Typhoon based on Risk Perception and CorrelationAnalysis between themAs can be seen from Table 5, risk perception in different dimensions has statistically significantimpact on active disaster response behavior, at p < 0.001. Further correlation analysis showed that thegreater the perceived risk, the more proactive measures the residents would take, which matches theprevious study. Matyas, Srinivasa and Cahyanto conducted a study on risk perception and evacuationdecisions of tourists in hurricane-affected areas, and found that the higher the risk perception,the stronger the evacuation willingness [15]. In addition, in the field of risk management, some scholarsfound that there is a significant positive correlation between risk perception and positive disasterresponse behavior [14,15].
Int. J. Environ. Res. Public Health 2020, 17, 7357 19 of 24In summary, hypothesis five that “participants’ risk perception significantly positively affectstheir motivation to respond to disasters” is accepted.3.5.6. Coupling Analysis of Knowledge, Information channels and Risk perceptionAs can be seen from the following scatter diagram Figure 9, first of all, the more informationacquisition channels and the less risk knowledge respondents have, the greater the risk perceptionof the overall impact of the typhoon; secondly, the fewer information access channels and less riskknowledge respondents have, the greater risk perceptions of property damage, health effects and lifethreats; finally, the more information access channels and the less risk knowledge respondents have,the higher their fear of typhoons.Int. J. Environ. Res. Public Health 2020, 17, x 21 of 26 previous study. Matyas, Srinivasa and Cahyanto conducted a study on risk perception and evacuation decisions of tourists in hurricane-affected areas, and found that the higher the risk perception, the stronger the evacuation willingness [15]. In addition, in the field of risk management, some scholars found that there is a significant positive correlation between risk perception and positive disaster response behavior [14,15]. In summary, hypothesis five that “participants’ risk perception significantly positively affects their motivation to respond to disasters” is accepted. Table 5. Knowledge, Information channels, risk perception & Active response to typhoon. Overall Impact Property Damage Health Effects Life Threat Fear Level F a r b F a r b F a r b F a r b F a r b K owledge Knowledge of typhoon signal 3.983 * −0.035 2.649 0.072 * 1.899 0.043 2.813 0.02 9.351 *** −0.079 * Understanding of typhoon prevention 2.622 * −0.077 * 14.332 *** −0.182 *** 9.886 *** −0.138 *** 8.068 *** −0.120 *** 3.692 ** −0.063 * Knowledge of typhoon preparedness 2.592 ** 0.061 9.098 *** −0.187 *** 14.833 *** −0.202 *** 13.129 *** −0.213 *** 3.286 ** 0.054 Knowledge (Weighted score) −0.037 −0.142 ** −0.146 ** −0.158 ** −0.065 * Numbers of information channels 3.396 ** 0.124 *** 6.485 *** −0.04 13.455 *** −0.112 *** 15.175 *** −0.139 *** 10.599 *** 0.246 *** Active response to typhoon 13.353 *** 0.202 ** 21.593 *** 0.277 ** 23.674 *** 0.259 ** 21.012 *** 0.247 ** 12.529 *** 0.204 ** * p<0.05, ** p<0.01, *** p<0.001, a ANOVA, b Correlation. 3.5.6. Coupling Analysis of Knowledge, Information channels and Risk perception As can be seen from the following scatter diagram Figure 9, first of all, the more information acquisition channels and the less risk knowledge respondents have, the greater the risk perception of the ov rall impact of th typhoon; s condly, the fewer inf rmation access channels nd l ss risk knowledge respondents have, the greater risk perceptions of property damage, health effects and life threats; finally, the more information access channels and the less risk knowledge respondents have, the higher their fear of typhoons. (a) (b) Int. J. Environ. Res. Public Health 2020, 17, x 22 of 26 (c) (d) (e) Figure 9. Coupling analysis of knowledge, information channels and risk perception. (a) Coupling analysis of knowledge, information channels and risk perception of the overall impact. (b) Coupling analysis of knowledge, information channels and risk perception of the property damage. (c) Coupling analysis of knowledge, information channels and risk perception of the health effect. (d) Coupling analysis of knowledge, information channels and risk perception of the life threat. (e) Coupling analysis of knowledge, information channels and risk perception of the fear level. 4. Conclusions and Prospects 4.1. Conclusions Although this research has certain shortcomings, it has some theoretical and practical significance in the field of risk perception. First of all, with questionnaires and statistical methods, such as hypothesis testing and correlation analyses, this paper studied such factors as demographic characteristics, personal or family circumstances, hedge knowledge, access to information and so on, are taken into consideration that may have an influence on public’s risk perception. (1) Judging from the average of the five dimensions of risk perception, the highest degree of public perception is the overall impact of typhoon risk (M = 4.14), followed by the degree of fear (M = 3.73) and then the property damage (M = 2.89), Life threat (M = 2.72) and health effect (M = 2.63). (2) The more knowledge and preventive measures for typhoon, the lower the risk perception, most of which are reflected in the three dimensions of property damage, health effect and life threat. This result is consistent with the study of Bettman and Park [11]. (3) The more typhoon information is obtained, the stronger the public’s overall risk perception and fear of the typhoon, and the weaker the risk perception of health effects and life threats. This result is consistent with the study of Wei et al. [12]. Figure 9. Coupling analysis of knowledge, information channels and risk perception.(a) Coupling analysis of knowledge, information channels and risk perception of the verall impact.(b) Coupling analysis of knowledge, information channels and risk perception of he prop rty damage.(c) Coupling analysis of kno ledge, information channels and risk perception of the health effect.(d) Coupling analysis of knowledge, information channels and risk perception of the life threat.(e) Coupling analysis of knowledge, information channels and risk perception of the fear level.
Int. J. Environ. Res. Public Health 2020, 17, 7357 20 of 24Table 5. Knowledge, Information channels, risk perception & Active response to typhoon.Overall Impact Property Damage Health Effects Life Threat Fear LevelF a r b F a r b F a r b F a r b F a r bKnowledgeKnowledge of typhoon signal 3.983 * −0.035 2.649 0.072 * 1.899 0.043 2.813 0.02 9.351 *** −0.079 *Understanding of typhoonprevention2.622 * −0.077 * 14.332 *** −0.182 *** 9.886 *** −0.138 *** 8.068 *** −0.120 *** 3.692 ** −0.063 *Knowledge of typhoonpreparedness2.592 ** 0.061 9.098 *** −0.187 *** 14.833 *** −0.202 *** 13.129 *** −0.213 *** 3.286 ** 0.054Knowledge (Weighted score) −0.037 −0.142 ** −0.146 ** −0.158 ** −0.065 *Numbers of information channels 3.396 ** 0.124 *** 6.485 *** −0.04 13.455 *** −0.112 *** 15.175 *** −0.139 *** 10.599 *** 0.246 ***Active response to typhoon 13.353 *** 0.202 ** 21.593 *** 0.277 ** 23.674 *** 0.259 ** 21.012 *** 0.247 ** 12.529 *** 0.204 *** p < 0.05, ** p < 0.01, *** p < 0.001, a ANOVA, b Correlation.
Int. J. Environ. Res. Public Health 2020, 17, 7357 21 of 244. Conclusions and Prospects4.1. ConclusionsAlthough this research has certain shortcomings, it has some theoretical and practical significance inthe field of risk perception. First of all, with questionnaires and statistical methods, such as hypothesistesting and correlation analyses, this paper studied such factors as demographic characteristics,personal or family circumstances, hedge knowledge, access to information and so on, are taken intoconsideration that may have an influence on public’s risk perception.(1) Judging from the average of the five dimensions of risk perception, the highest degree of publicperception is the overall impact of typhoon risk (M = 4.14), followed by the degree of fear (M = 3.73)and then the property damage (M = 2.89), Life threat (M = 2.72) and health effect (M = 2.63).(2) The more knowledge and preventive measures for typhoon, the lower the risk perception,most of which are reflected in the three dimensions of property damage, health effect and life threat.This result is consistent with the study of Bettman and Park [11].(3) The more typhoon information is obtained, the stronger the public’s overall risk perceptionand fear of the typhoon, and the weaker the risk perception of health effects and life threats. This resultis consistent with the study of Wei et al. [12].(4) The stronger the risk perception, the more positive disaster response behaviors the peopleadopt, which is consistent with the researches of Riad et al. and Matyas et al. [14,15]. Furthermore,the positivity of the people should be mainly reflected in the prevention before the disaster.(5) There are significant differences in the perception of typhoon risk among people of differentage groups. The risk perception of the typhoon of the 25–44 age group is significantly higher than thatof the other groups. This result is reflected in the five dimensions of risk perception. The age groupof over 65 years of age has a significantly lower risk perception of property damage, life threats andhealth effects than the other groups. From the perspective of the correlation between age and riskperception, age has a significant negative correlation with the risk perception of property damage,health effects and life threats.(6) Respondents with different education have significant differences in risk perception of the fourdimensions of property damage, life threat, health effect and degree of fear. The risk perception ofrespondents in the high school education in the three dimensions of property damage, life threats,and health effects is significantly higher than those of the other groups; respondents with primaryschool education and below have a significantly lower risk perception in these three dimensionsthan other groups, but their fear levels are significantly higher than other groups. In terms of thecorrelation between education level and risk perception, there is a significant positive correlationbetween education level and risk perception of property damage, health impact and life threats.(7) Respondents of different occupations have significant differences in typhoon risk perception.From the perspective of overall impact, legislators, government officials, community leaders,business leaders and managers, and handicrafts-man have higher risk perceptions, and serviceand sales staff and students have lower risk perceptions; the risk perception in the three dimensions ofproperty damage, health effect and life threats of the three occupations of professionals, technicians andsupport professionals, and handicrafts workers is higher and of the unemployed is lower in the degreeof fear. People of other occupations are of the highest degree of fear and students are of the lowestdegree. In terms of the correlation between health status and risk perception, although there is asignificant correlation, the correlation coefficient is low and can be ignored.(8) Respondents with different personal and family situations also showed significant differencesin typhoon risk perception from various dimensions. In terms of living floors, respondents livingon the ground floor or below have higher risk perception; in terms of health, respondents with poorphysical condition have higher risk perception; in terms of cities of residence, respondents living inMacao have the lowest risk perception, and respondents who live in areas vulnerable to typhoons havea higher risk perception; in terms of length of stay, the longer the residents stay, the higher their risk
Int. J. Environ. Res. Public Health 2020, 17, 7357 22 of 24perception is in the three dimensions of property damage, health effect and life threat, and the lowerthe degree of fear is; in terms of the structure of family members, respondents with children under14 years of age, elderly people over 65 years of age or members with reduced mobility have a strongerrisk perception; finally, in terms of personal monthly income, the higher the income, the stronger therisk perception of property damage, health effects and life threats.Secondly, this research is the first to analyze risk perception from five aspects: overall impact,property damage, health impact, life threat, and degree of fear. Furthermore, the results revealed thatalthough there is a high degree of correlation between various aspects of risk perception, when it comesto the relationship between them and other factors (such as personal circumstances, risk knowledge,or typhoon information acquisition channels, etc.), there are still some differences.Thirdly, the existing research on risk perception, the demographic characteristics involved onlyincludes basic information such as gender and age, while this research involves more further personalconditions such as residential area, residential floor and family member organizational structure.In comparison, the research on the factors affecting risk perception has a wider coverage.Fourthly, the more information acquisition channels and the less risk knowledge respondentshave, the greater the risk perception of the overall impact of the typhoon; the fewer informationaccess channels and less risk knowledge respondents have, the greater the risk perceptions of propertydamage, health effects and life threats; the more information access channels and the less risk knowledgerespondents have, the higher their fear of typhoons.Finally, although the study uses Macao as the research area, most of the research results areconsistent with previous research results. Therefore, the results of this study can be generalized andapplied to most coastal cities or island cities.4.2. ProspectsThis research only studies the influencing factors of public risk perception of typhoon.Future research can be extended from the following three aspects:Firstly, fully consider the role of risk information dissemination on public risk perception. With thehelp of big data platform, we can obtain massive risk information and public preference data ofdiversified communication media after the occurrence of emergencies, explore the characteristics ofpublic information demand and measure the psychological gap caused by the difference betweenpublic demand information and risk information. On this basis, with the help of behavioral decisiontheory and psychological theory, the public’s risk perception is portrayed, and according to the public’sdifferent preferences of the media, the change law of risk perception can be studied.Secondly, the life cycle factors of emergencies can be introduced in the description of risk perception.By refining the life cycle of emergencies, scholars can conduct a phased study of public risk perception,predict the public’s psychological expectations based on the different phases of the event situation,and refine the influencing factors and mechanism of public risk perception, then measure publicpsychological perception at various stages based on this. Further, by considering the influencingfactors of the evolution of different stages of the life cycle of emergencies, the evolution of publicpsychological perception between different stages can be studied.Thirdly, on the one hand, follow-up research on risk perception can take the system trust intoconsideration, including expert trust, market trust, government trust and media trust; on the otherhand, interest-related factors may also affect individual risk perception. Benefits include not onlypositive benefits but also negative benefits, and perspective of long-term or short-term benefits canbe considered. For example, in the results of this article, it is found that some respondents who havelived in Macao for 5–10 years have stronger risk perception, which may be related to the permanentresidence policy of Macao, which may be considered as an interest-related factor.Author Contributions: Conceptualization, Y.S., S.L. and X.Z.; methodology, Y.S., S.L. and X.Z.; validation, Y.S.,S.L., X.Z., H.X. and J.Z.; formal analysis, Y.S., S.L. and X.Z.; investigation, S.L., X.Z., Y.S., H.X. and J.Z.; data curation,Y.S., X.Z., S.L., H.X. and J.Z.; writing—original draft preparation, Y.S.; writing—review and editing, Y.S., S.L. and
Int. J. Environ. Res. Public Health 2020, 17, 7357 23 of 24X.Z.; visualization, Y.S.; supervision, K.P.I., S.L. and X.Z.; project administration, K.P.I. and S.L.; funding acquisition,K.P.I. and S.L. All authors have read and agreed to the published version of the manuscript.Funding: This research was funded by National Natural Science Foundation of China, grant number 71861167002;Macao Science and Technology Development Fund, grant number 0049/2018/AFJ and Macao Science andTechnology Development Fund, grant number 0164/2019/A3.Acknowledgments: We thank Macau General Neighbourhood Association (União Geral das Associações dosMoradores de Macau) for its help in the survey.Conflicts of Interest: The authors declare no conflict of interest.References1. PMF IAS. Tropical Cyclones: Favorable Conditions for Formation, Stages of Formation & Structure.2016. Available online: https://www.pmfias.com/tropical-cyclones-favorable-conditions-tropical-cycloneformation/ (accessed on 6 January 2016).2. Mei, W.; Xie, S. Intensification of landfalling typhoons over the northwest Pacific since the late 1970s.Nat. Geosci. 2016, 9, 753–759. [CrossRef]3. Cohen, B.L. Public perception versus results of scientific risk analysis. Reliab. Eng. Syst. Saf. 1998, 59, 101–105.[CrossRef]4. Rogers, G.O. The dynamics of risk perception: How does perceived risk respond to risk events? Risk Anal.1997, 17, 745–757. [CrossRef]5. Pidgeon, N. Risk assessment, risk values and the social science programme: Why we do need risk perceptionresearch. Reliab. Eng. Syst. Saf. 1998, 59, 5–15. [CrossRef]6. Slovic, P. Perception of risk. Science 1987, 236, 280–285. [CrossRef]7. Sjöberg, L.; Moen, B.E.; Rundmo, T. Explaining risk perception. An evaluation of the psychometric paradigmin risk perception research. Rotunde Publikasjoner 2004, 84, 55–76.8. Lai, J.; Tao, J. Perception of environmental hazards in Hong Kong Chinese. Risk Anal. 2003, 23, 669–684.9. Spence, H.E.; Engel, J.F.; Blackwell, R.D. Perceived risk in mail-order and retail store buying. J. Mark. Res.1970, 7, 364. [CrossRef]10. Jackson, E.L. Response to Earthquake Hazard: The West Coast of North America. Environ. Behav. Behav.1981, 13, 387–416. [CrossRef]11. Bettman, J.R.; Whan, P.C. Effects of prior knowledge and experience and phase of the choice process onconsumer decision processes: A protocol analysis. J. Consum. Res. 1980, 7, 3. [CrossRef]12. Peacock, W.G.; Brody, S.D.; Highfield, W. Hurricane risk perceptions among florida’s single familyhomeowners. Landsc. Urban Plan. 2005, 73, 120–135. [CrossRef]13. Wei, J.; Wang, F.; Zhao, D. A risk perception model: Simulating public response to news reports in China.Inf. Res. 2012, 17, 519.14. Riad, J.K.; Norris, F.H.; Ruback, R.B. Predicting evacuation in two major disasters: Risk perception,social influence, and access to resources1. J. Appl. Soc. Psychol. 2006, 29, 918–934. [CrossRef]15. Matyas, C.; Srinivasan, S.; Cahyanto, I.; Thapa, B.; Pennington-Gray, L.; Villegas, J. Risk perception andevacuation decisions of florida tourists under hurricane threats: A stated preference analysis. Nat. Hazards2011, 59, 871–890.16. Santos-Reyes, J.; Santos-Reyes, G.; Gouzeva, T.; Velazquez-Martinez, D. Schoolchildren’s earthquakeknowledge, preparedness, and risk perception of a seismic-prone region of Mexico. Hum. Ecol. Risk Assess.2017, 23, 494–507. [CrossRef]17. Wachinger, G.; Renn, O.; Begg, C.; Kuhlicke, C. The risk perception paradox-implications for governance andcommunication of natural hazards. Risk Anal. 2013, 33, 1049–1065. [CrossRef]18. Bodoque, J.M.; Díez-Herrero, A.; Amérigo, M.; García, J.A.; Olcina, J. Enhancing flash flood risk perceptionand awareness of mitigation actions through risk communication: A pre-post survey design. J. Hydrol. 2019,568, 769–779. [CrossRef]19. Henry, L.; Takagi, H.; Viet, N.T.; The, N.N. Applicability of Stormwater—River Flood Combined Modelin Low-Lying Delta: Disaster Survey in Hoi an and Model Assessment of 2017 Typhoon Damrey. InProceedings of the APAC 2019 International Conference on Asian and Pacific Coasts, Hanoi, Vietnam,25–28 September 2019; Trung Viet, N., Xiping, D., Thanh Tung, T., Eds.; Springer: Singapore, 2020.
